Matching Networks for One Shot Learning

Learning from a few examples remains a key challenge in machine learning. Despite recent advances in important domains such as vision and language, the standard supervised deep learning paradigm does not offer a satisfactory solution for learning new concepts rapidly from little data. In this work, we employ ideas from metric learning based on deep neural features and from recent advances that augment neural networks with external memories. Our framework learns a network that maps a small labelled support set and an unlabelled example to its label, obviating the need for fine-tuning to adapt to new class types. We then define one-shot learning problems on vision (using Omniglot, ImageNet) and language tasks. Our algorithm improves one-shot accuracy on ImageNet from 87.6% to 93.2% and from 88.0% to 93.8% on Omniglot compared to competing approaches. We also demonstrate the usefulness of the same model on language modeling by introducing a one-shot task on the Penn Treebank.

PDF Abstract NeurIPS 2016 PDF NeurIPS 2016 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Few-Shot Image Classification Meta-Dataset Matching Networks Accuracy 56.247 # 19
Few-Shot Image Classification Meta-Dataset Rank Matching Networks Mean Rank 10.5 # 11
Few-Shot Image Classification Mini-Imagenet 5-way (1-shot) Matching Nets (Cosine Matching Fn) Accuracy 46.6 # 103
Few-Shot Image Classification Mini-Imagenet 5-way (5-shot) Matching Nets (Cosine Matching Fn) Accuracy 60 # 94
Few-Shot Image Classification OMNIGLOT - 1-Shot, 20-way Matching Nets Accuracy 93.8% # 15
Few-Shot Image Classification OMNIGLOT - 1-Shot, 5-way Matching Nets Accuracy 98.1 # 14
Few-Shot Image Classification OMNIGLOT - 5-Shot, 20-way Matching Nets Accuracy 98.5% # 13
Few-Shot Image Classification OMNIGLOT - 5-Shot, 5-way Matching Nets Accuracy 98.9 # 16
Few-Shot Image Classification Stanford Cars 5-way (1-shot) Matching Nets FCE++ Accuracy 34.80 # 6
Few-Shot Image Classification Stanford Cars 5-way (5-shot) Matching Nets FCE++ Accuracy 44.70 # 6

Results from Other Papers


Task Dataset Model Metric Name Metric Value Rank Source Paper Compare
Few-Shot Image Classification Mini-ImageNet-CUB 5-way (1-shot) MatchingNet (Vinyals et al., 2016) Accuracy 45.59 # 5
Few-Shot Image Classification Stanford Dogs 5-way (5-shot) Matching Nets FCE++ Accuracy 47.50 # 6

Methods


No methods listed for this paper. Add relevant methods here